IRC logs for #openttd on OFTC at 2017-07-09
⏴ go to previous day
00:43:32 *** sla_ro|master has joined #openttd
02:37:29 *** Hiddenfunstuff has quit IRC
02:52:59 *** HerzogDeXtEr has joined #openttd
05:48:28 *** techmagus has joined #openttd
05:54:01 *** Biolunar has joined #openttd
07:05:25 *** Supercheese has joined #openttd
07:47:22 *** Alberth has joined #openttd
07:47:22 *** ChanServ sets mode: +o Alberth
08:22:59 *** Progman has joined #openttd
08:49:33 *** Deactivated has joined #openttd
09:39:11 *** andythenorth has joined #openttd
09:59:48 *** sla_ro|master has joined #openttd
10:55:08 *** sim-al2 has joined #openttd
11:01:35 *** HerzogDeXtEr has joined #openttd
11:13:14 * andythenorth wants to rework Iron Horse
11:13:18 <andythenorth> but must FIRS :(
11:22:02 *** FLHerne has joined #openttd
11:23:30 *** Wormnest has joined #openttd
11:26:42 *** mindlesstux has joined #openttd
11:28:21 <andythenorth> and then manually assign them
11:28:33 <andythenorth> repeating map colours isn’t terrible for industries
11:33:21 <andythenorth> sometimes there’s no automated solution :P
11:34:09 <Alberth> in general, it worked quite nicely, just the greens and blue watery colour were not so great
11:34:29 <Alberth> you tried just leaving those out?
11:35:34 <Alberth> biggest issue is that colours are completely different from default industries, and from previous FIRSes :)
11:35:40 <andythenorth> is that an issue?
11:35:57 <andythenorth> at first I liked the difference, because it’s new
11:36:03 <andythenorth> but then in my test games I got annoyed
11:36:09 <andythenorth> “why are you hiding the coal mines"
11:36:23 <Alberth> I failed to recognize anything :)
11:36:35 <Alberth> but it's a matter of time to adjust
11:37:07 <andythenorth> I might set manual colours, with a guard against the map problem
11:37:21 <andythenorth> I can probably check if colour is in first 64 or so of what Frosch generated
11:37:28 <andythenorth> with some special case for sea
11:38:17 <Alberth> if a colour that you used before is in the first 64, there isn't much reason not to keep that as-is
11:38:56 <Alberth> eg black coal mines still work then
11:40:39 <Alberth> on the other hand, eg brown iron-ore mines in default are hard to see, no reason to keep that
11:41:19 <andythenorth> looks like the algorithm rules out black :)
11:41:23 <andythenorth> can’t see it in the table
11:43:27 <Alberth> 4.67 is x202020 which is also dark
11:44:20 <Alberth> technically, you could get a clash with "outside map", but that's a non-issue imho
11:50:34 *** Stimrol has joined #openttd
11:57:37 <LordAro> Alberth: i do believe i can remove the CDECL from the function calls as well, given it's related to the C calling convention
12:02:12 <Alberth> hmm, stupid MS inventing own keywords
12:03:02 <Alberth> I believe there was also something related to preserving the function name, ie skip name mangling
12:03:19 <Alberth> but euhm, I don't know, never used Windows
12:03:42 <Alberth> make it a separate patch at least, perhaps another dev can help you here
12:09:45 <LordAro> hmm, SmallVector::Clear is weird. i feel like it should be freeing memory, but it just resets a counter
12:10:55 <LordAro> i'm not actually all that sure it has any advantages over std::vector anyway
12:12:58 <Alberth> SmallVector pre-dates vector by a decade or so
12:13:15 <LordAro> vector isn't new to C++...
12:13:35 <Alberth> ok, less than a decade then :p
12:13:52 <LordAro> regardless, separate patch
12:14:03 <Alberth> code started in C, where you don't have templates
12:15:07 *** frosch123 has joined #openttd
12:15:18 <Alberth> iirc the biggest advantage was that you could add a value without copying it, like push_back does
12:18:10 <frosch123> andythenorth: yes, the algorithms banishes black
12:18:42 <frosch123> the first version had many dark colours, which where hard to distinguish, so i penalised dark colours
12:18:55 <frosch123> ofc there is space for *one* dark colour :p
12:19:02 <Alberth> argh, how I hate lua with its silent replacement by 'nil'
12:19:19 <frosch123> what are you doing with lua?
12:19:48 <Alberth> writing a property reader/writer, or rather attempting to
12:19:51 <frosch123> i encounter it a lot recently
12:20:11 <frosch123> like at work, in f and now here
12:20:11 <Alberth> corsix-th is full of it
12:20:39 <Alberth> maybe I should refuse to use it
12:22:21 <frosch123> LordAro: the main point of SmallVector is that it gives the advantage of emplace long before c++11 existed
12:23:15 <adf88> I have a few small patches
12:23:29 <adf88> would you like to take a look?
12:23:47 <adf88> these are straightforward fixes
12:24:05 <LordAro> frosch123: aye, probably
12:24:51 <frosch123> i saw them in the log
12:25:16 <frosch123> i need to find a working copy without modifies :p
12:25:51 <frosch123> maybe i should look into using a better svn client
12:28:43 <Alberth> that fails at least for empty directories
12:29:30 <frosch123> most cross-vcs things fail with svn:keywords
12:29:36 <frosch123> maybe we should remove them
12:29:54 <Alberth> not to mention git fails on big file moves
12:30:09 <LordAro> it's been getting better
12:30:21 <LordAro> although you do have to massage it at times, i'll admit
12:31:06 <Alberth> RB wanted svn keywords for identification of a file in case we ran into one in the wild
12:44:05 <andythenorth> top 64 contrasting map colours?
12:47:05 <andythenorth> and they’re mostly orange or pink
12:47:45 <frosch123> you banned all of green and half of violet
12:48:00 <frosch123> i guess top 32 might work
12:48:05 <frosch123> but anything more is too much
12:49:27 <andythenorth> I’ll do the work to shuffle them into 64
12:49:38 <andythenorth> and if that’s lame, I’ll do it again to fit them into 32
12:54:21 <andythenorth> frosch123: can the algorithm force-include colours?
12:54:28 <andythenorth> not essential, just curious
12:55:08 <frosch123> i essentially force-included the map colours and black
12:55:17 <frosch123> that's why the algorithm stays away of them
12:55:34 <frosch123> there is no difference between "force-include" and "ban"
12:55:54 <frosch123> in the former case you just put them manually into front
12:58:50 <frosch123> it's no clever algorithm :p just greed
12:59:22 <frosch123> you can manually pick some colors and then auto-assign the rest just fine
13:00:49 <andythenorth> I am using the list for a compile-time guard
13:00:53 <andythenorth> with manual assignment
13:27:14 <frosch123> does it make sense to add those colors to the economy graphs?
13:27:54 <andythenorth> might be too much colour
13:27:57 <andythenorth> want to try? o_O
13:28:03 <frosch123> not to fill the whole boxes, that would make it unreadable
13:28:11 <frosch123> but maybe some square somewhere
13:34:13 <andythenorth> pleased the map colours are fixed :)
13:36:13 <LordAro> (this is with FileList restored to its original state)
13:37:14 <LordAro> ah yes, it's not using the const version of Filelist::Get
13:37:25 <LordAro> for reasons i can't tell
13:37:42 <Alberth> line 5 looks wrong, imho
13:38:41 *** Hiddenfunstuff has joined #openttd
13:38:42 <LordAro> i've got another branch for that
13:39:10 <Alberth> isn't it just that it has no type for CompareFiosItems ?
13:40:02 <Alberth> what happens if you declare it?
13:40:32 <LordAro> oh, i forgot about the header declaration
13:45:24 <Alberth> hmm, weird indeed, FileList looks ok
13:45:55 <Alberth> perhaps it gets confused due to the additional pointer, which is not const in both cases
13:46:59 <Alberth> tbh I have yet to find a reason why the pointer should be const in these cases
13:47:17 <Alberth> who cares what the caller does with the pointer?
13:47:54 <LordAro> i'd imagine it just cares that they are the same, rather than making any distinction between const/nonconst
13:49:09 <LordAro> that said, if i comment out the non-const Get function, it explodes into templatehell
13:52:01 <Alberth> oh, perhaps swapping during sort?
13:54:25 <LordAro> i think that's what the error messages are saying :)
13:56:35 * LordAro adds more code duplication in the form of a nonconst End
13:57:19 <frosch123> i wonder why no c++xx has added a keyword for const/nonconst c&p
14:00:21 <Alberth> not sure what c&p means, but perhaps not enough added value?
14:02:13 <adf88> void this_is_cost_and_non_const_method() const_or_non_conts
14:02:19 <frosch123> i guess it does not work well with templated parameters
14:02:34 <adf88> this->something(); // const or non-const
14:03:30 <frosch123> but for "autoconst Foo& GetFoo() autoconst { return this->foo; }" it would work
14:03:43 <Alberth> in general both implementations can be very different
14:03:55 <frosch123> not so much if it needs to distinguish iterator and const_iterator
14:04:27 <Alberth> maybe they hope to make the compiler smart enough to recognize const
14:05:58 <Alberth> always nice to find branches that are just old copies of 'master' :p
14:06:40 <adf88> auto iter = container->begin();
14:07:28 <frosch123> well, but does "auto" work for function parameters and return values?
14:08:39 <adf88> sorry, I just got in the middle and I don't know what you are trying to achieve :p
14:08:57 <frosch123> ah well, perhaps boost will discover some magic for "boost::if_const<const_iterator, iterator>" :p
14:08:58 <Alberth> making a spec for the next C++xx :p
14:09:11 <adf88> but I would say: create a function template, it will accept "auto" var
14:09:48 <adf88> T will be either 'const_iterator' or 'iterator'
14:10:37 <Alberth> git rename guesses are hilarious :p "rename CorsixTH/{Lua/command.lua => Graphics/file_mapping.txt} (87%)"
14:11:35 <frosch123> reminds me of gettext fuzzy matching
14:12:04 <Alberth> rename MapEdit/Src/app.cpp => CorsixTH/Lua/entities/grim_reaper.lua
14:20:33 <LordAro> frosch123: auto works in C++17, i think
14:21:18 <LordAro> lambdas are getting fun though - i made a thing a few months ago that was signficantly simpler with a load of top-level lambdas instead of "proper" functions
14:30:33 <Eddi|zuHause> speaking of templates, i had a go with C++ a while back, trying out a function like "template<typename T> T f(T param)", but how do you instantiate the template properly, if the function is not called in this .cpp file, but in another one? i had to manually put instantiations like "int f(int)" "byte f(byte)" etc...
14:31:41 <frosch123> usually you put the implementation into the header
14:31:56 <frosch123> it's were rare to do the instantiation in a cpp file
14:31:58 <Eddi|zuHause> but that would duplicate the function?
14:32:10 <frosch123> the linker sorts that out
14:32:56 <frosch123> the linker generally sorts out duplication of implicitly declared thigns
14:32:58 <Eddi|zuHause> doesn't really scream "the right" solution to me
14:33:18 <frosch123> like default constructor, copy constructor, destructor, virtual table, run time type info, ...
14:34:01 <frosch123> Eddi|zuHause: the point of a template is that it works with many types, not just with the 5 you thought of when writing the template
14:35:38 <Eddi|zuHause> in particular, i was cleaning up a file where the person just copy-pasted the same function 20 times, just swapping out int32/int16/uint16/bool/...
14:38:24 <Alberth> they solved that linker problem around 20 years ago
14:39:23 <frosch123> i really like languages which do not have "unsigned" stuff
14:40:16 <Eddi|zuHause> i can't do everything in python :p
14:42:24 <Eddi|zuHause> in this particular case this was some network interface code, where the type had to match the underlying (and fixed) format
14:43:07 <Alberth> we over-use 'unsigned' in openttd, you should only use it for collections of bits, mostly
14:43:14 <Eddi|zuHause> i also had "great fun" trying to reimplement that interface in python, with the c_types module
14:43:56 <Eddi|zuHause> Alberth: as soon as you use "-" on a type, it should be signed
14:44:32 <Alberth> even before, eg population or length should be signed too
14:45:21 <Alberth> ie "non-negative values only" is not sufficient for using unsigned
14:46:07 <Alberth> shit starts to drop when you are using the full width of integers :p
14:46:32 <Eddi|zuHause> well, the extended range really matters if you have to weigh in whether you use 8 or 16 bit, but once you're using 32, 64 or more bits, it's almost irrelevant
14:47:25 *** gelignite has joined #openttd
14:48:20 <Eddi|zuHause> i've come across an "unsigned 64 bit integer" which has the meaning "100ns units since 1st january 1601"
14:48:52 <Alberth> shorter integers is probably also overengineering, just use int everywhere
14:49:47 <Alberth> of course, network IO and file formats have different considerations here
14:49:57 <Eddi|zuHause> that turns out to be quite hard to remember if your language has "INT" meaning "16 bits"
14:50:20 <frosch123> @calc 2**64/(10**7 *3600*24*365)
14:50:20 <DorpsGek> frosch123: 58494.2417355
14:50:46 <frosch123> Eddi|zuHause: quite conservative choice, 10ns would also last till 5800"
14:50:47 <Alberth> longer than expected :p
14:51:17 <frosch123> but i see that they did not pick 1ns
14:51:19 <Eddi|zuHause> frosch123: maybe someone thought "ah well, we won't need better resolution"
14:52:02 <frosch123> i think every time format should have an addition field for number of leap-seconds included
14:53:01 <Eddi|zuHause> just turn off the computer for 24h on days with leap seconds
14:53:40 <Eddi|zuHause> that is apparently the default solution for SAP systems with summer/winter time shifts :p
14:54:48 <Alberth> just use GMT-relative time
15:16:44 <Eddi|zuHause> LordAro: i feel like i read a simplified excerpt from that recently
15:18:26 <Eddi|zuHause> i'm not sure my processes are time-critical enough to worry about leap seconds, though
15:19:37 <Eddi|zuHause> plus, they won't be running over night, probably
15:31:52 <frosch123> yeah, first make them crash less often :p
15:32:38 <Eddi|zuHause> did you mean "explode"? :p
15:33:17 <adf88> I wish they had do this leap second thing otherwise
15:33:17 <adf88> one second in GMT time doesn't have to be equivalent to a physical second
15:33:17 <adf88> virtual GMT seconds could last a little longer or a little slower then a physical second
15:33:17 <adf88> the central committee would be fine-tuning their central clock so it slowly drifts to compress the upcoming changes in Earth movement around sun
15:33:18 <adf88> computers would desynchronize slowly from the master clock, BUT THEY DO SO ANYWAY
15:33:18 <adf88> in it's time, they would safely resynchronize with the master clock, THEY HAVE TO DO IT ANYWAY
15:33:41 <Eddi|zuHause> i do PLC programming, which usually means there's an actual machine attached to it :p
15:34:48 <Eddi|zuHause> adf88: google does something like that, on days with leap seconds they stretch out the seconds over the day ever so slightly
15:36:51 <adf88> for high performance time calculations you don't use GMT anyway so GMT doesn't have to be "scientific"
15:40:11 <Eddi|zuHause> adf88: that's a dangerous claim
15:41:18 <adf88> i meant really high like GPS stuff etc
15:51:42 *** mindlesstux has joined #openttd
15:53:26 <adf88> I heard (and It seems natural) that for scientific calculations like those in space, UTC is not usefull (because of the leap), they use other timers that are not affected by Earh movement around Sun
15:55:13 <Eddi|zuHause> adf88: that's explained in above link
16:00:17 <frosch123> yeah, in space they use parallax seconds and light years :p
16:00:41 <Eddi|zuHause> frosch123: yes, those are the best time units :p
16:17:49 <adf88> I remember once when polish handball representation was playing with Norway
16:17:49 <adf88> 15 seconds to the end, we are loosing, Bogdan Wenta (coach) takes a break and says to the team:
16:17:49 <adf88> "guys, easy, there's a lot of time"
16:17:49 <adf88> since then, 1 venta = 15 seconds :D
16:22:49 <Eddi|zuHause> depends on whether you can partition those 15 seconds into multiple attack sequences (e.g. by american-football-like timeouts)
16:34:21 *** andythenorth has joined #openttd
16:35:04 *** sla_ro|master has joined #openttd
17:22:31 <andythenorth> is ottd reddit any good?
17:22:50 <frosch123> don't ask anyone from btpro :p
17:23:16 <frosch123> there is some drama which i do not quite follow
17:26:37 <andythenorth> so what icon for Recyclables cargo?
17:26:48 <andythenorth> I tried the 3 arrows in a circle for ‘recycle'
17:27:25 <frosch123> just wanted to suggest that
17:27:36 <andythenorth> arrows don’t fit the number of pixels
17:31:16 <andythenorth> due to the grid, one arrow must be much larger than the other two
17:32:43 * andythenorth wonders about a 32bpp UI with 8bpp game sprites
17:33:09 <frosch123> draw 2x cargo icons?
17:33:19 <frosch123> you were using 2x gui, right?
17:37:25 <frosch123> vehicle body is hard though, if you also have vehicles
17:40:39 <andythenorth> drawing icons is totally different to drawing sprites :P
17:40:51 <frosch123> many of your icons try 3d
17:41:00 <frosch123> 2d may be easier with that little pixels
17:42:42 <andythenorth> Foobar and Dan drew most of the other icons
18:21:10 *** ChanServ sets mode: +v tokai
19:56:40 *** FLHerne has joined #openttd
20:46:24 <andythenorth> it bothers me there are two different styles of mineral pile
20:47:09 *** FLHerne has joined #openttd
20:48:57 <frosch123> i think it's reasonable
20:49:38 <frosch123> i don't think it is any worse than the other small icons
20:49:49 <frosch123> most icons have the problem of too few pixels
20:50:03 * andythenorth wonders if we should change them
20:53:34 <andythenorth> Alberth: does FIRS 2 default to Extreme for you? o_O
20:53:50 <andythenorth> I get Temperate Basic
20:56:45 *** ChanServ sets mode: +v Alberth
21:06:30 <frosch123> hmm, i can get the map colours into the cargo graphs
21:06:46 <frosch123> but there is no equivalent of showing a partial image
21:06:58 <frosch123> so cargo icons would only work with one icon per image
21:08:43 <Eddi|zuHause> andythenorth: have you ever considered having the default be "Basic (climate-dependent)" and then additional economies for "Basic (Arctic)", "Basic (Temperate)", "Basic (Tropic)", others...
21:09:19 <frosch123> Eddi|zuHause: yesterday we tried to make a point that you can play any economy in any climate :p
21:09:31 <Eddi|zuHause> you could still do that with this
21:09:45 <Eddi|zuHause> but you bring the most diversity to people that never change parameters
21:11:04 <Eddi|zuHause> and you can then put a sentence about the climate-independentness of the climate-economy in the description
21:11:43 <Alberth> andy, nah, I use BB as guide :)
21:16:35 <Alberth> the OP was complaining FIRS was too complicated
21:26:18 <Eddi|zuHause> that's not a problem, if you ask enough people, you always find people who have these extreme opinions
21:27:02 <Eddi|zuHause> meaning it's simultaneously too complicated and not nearly complicated enough
21:27:38 <andythenorth> I don’t know what to say to over-complicated
21:27:46 <andythenorth> it’s 19 industries versus 12
21:27:56 <andythenorth> temperate / temperate basic
21:28:12 <andythenorth> or temperate basic / temperate :P
21:29:22 <andythenorth> frosch123: I could slice out the cargo icons with PIL :P
21:30:25 <frosch123> hmm, colors do not seem to work either
21:30:45 <frosch123> maybe would require templating the css as well
21:31:24 <andythenorth> dunno if it’s worthwhile :)
21:32:24 <andythenorth> can you remember why we made primaries green and secondaries red?
21:32:34 <andythenorth> I am considering making them all variations of blue or so
21:33:41 <frosch123> we tried different colors
21:33:56 <frosch123> there were several version
21:34:00 <frosch123> i like the current ones
21:35:10 <frosch123> in particular i like green for primary
21:35:17 <frosch123> because they produce on their own
21:36:02 <frosch123> blue is generally a boring color, so making everything blueish would look sad
21:36:04 <andythenorth> seems we tried quite a few options
21:36:15 <andythenorth> there was yellow at some point
21:36:23 <andythenorth> and we distinguished extractive / organic
21:36:41 <frosch123> i can see why that was removed :p
21:36:45 * andythenorth leaves them alone
21:36:48 <frosch123> sounds pretty pointless
21:38:03 <frosch123> oh, maybe the css is working against my colors
21:47:24 *** Wormnest has joined #openttd
21:51:16 <andythenorth> my to-do list says “cargo weights seem to be arbitrary, fix them"
21:51:17 <andythenorth> file:///Users/andy2/Documents/OTTD_graphics/FIRS/firs/docs/html/code_reference.html#cargos
21:51:43 <andythenorth> local links never work for you lot :P
21:54:26 <andythenorth> how much should 1t of alcohol weigh then?
21:54:55 <andythenorth> or do the bottles make it heavier?
21:55:04 <frosch123> i think you deprecated that task by making most cargos use "1 ton" insead of "1 item/bag/sack"
21:55:30 <andythenorth> should I put the units in the table?
21:55:38 <andythenorth> all 1 ton are at least easy :P
21:56:44 * andythenorth wonders why the svgs don’t work for me from there
21:57:28 <andythenorth> tried all three browsers :)
21:59:26 <frosch123> devs.openttd.org/~frosch/firs/html5/wip5.diff <- render them yourself :)
21:59:36 <frosch123> the Makefile change may be useful in any case
21:59:57 <frosch123> also the ratio thing makes some economies look better, though it may become a economy-specific setting
22:04:38 <andythenorth> it’s interesting frosch123
22:08:03 <andythenorth> dunno if the svg css permits that
22:17:54 <andythenorth> the ratio does seem to be an improvement
22:24:02 <andythenorth> beer is denser than water :o
22:31:10 <frosch123> so if you let beer settle for a while: there is alcohol at the top, then water, then the rest?
22:33:16 <andythenorth> weight of livestock :P
22:33:21 <andythenorth> chickens, sheep or cows?
22:36:25 <andythenorth> seems we already set it to weight of ponies :P
22:38:58 <supermop_home> apparently melb is building out some suburban line by 3 more stations
22:38:59 <andythenorth> FIRS 3 creeps closer to the end
22:39:14 <andythenorth> bothers me I can’t re-order the economies
22:39:19 <andythenorth> Extreme should be last
22:39:31 <andythenorth> maybe I just break savegames, that happens anyway
22:41:24 <frosch123> i think you can reorder them
22:41:29 <frosch123> without breaking savegames
22:41:52 <frosch123> add "+16" to the values until they are in the order you want
22:42:09 <frosch123> then do "param_foobar = param_foobar & 0xF" at the start of the grf
22:42:56 <andythenorth> is too late in day for me to figure out what that does :)
22:43:16 <andythenorth> I have elaborate but flawed code already to try and avoid breaking savegames
22:43:23 <andythenorth> it does nothing useful :)
22:44:02 <frosch123> essentially it make the combobox choose paired-values (economy, sortorder)
22:44:15 <frosch123> lower nibble is economy, upper nibble is sortorder
22:44:47 <andythenorth> I need to implement that tomorrow :)
22:45:44 *** sim-al2 has joined #openttd
22:48:33 *** andythenorth has left #openttd
23:11:18 *** FLHerne has joined #openttd
23:22:03 *** andythenorth has joined #openttd
23:22:11 <andythenorth> Manufacturing Supplies has no special behaviour
23:22:15 <andythenorth> unlike ENSP and FMSP
23:22:23 <andythenorth> it should be removed / renamed / rethought
23:24:16 <andythenorth> it’s consumed at 9 industries
23:24:21 <andythenorth> 6 of them it’s just packaging
23:24:39 <andythenorth> the other I have NFI
23:25:25 <andythenorth> crates of boxes? o_O
23:25:36 <supermop_home> boxes of crates
23:25:50 <supermop_home> jars of bottles
23:25:51 <andythenorth> oh that’s just cardboard :P
23:26:17 <andythenorth> texitle mill should _produce_ MNSP, not consume it :P
23:26:28 <andythenorth> ‘Packaging’ then?
23:26:41 <andythenorth> and remove it from Furniture Factory and Textile Mill?
23:28:55 <andythenorth> IRL, it should also cover the common industrial chemicals like salt
23:29:03 <andythenorth> but that didn’t happen
continue to next day ⏵
